Face Identification on IJB-A (1:N Protocol)
99.19Rank-1ArcFace
Evaluation results
| Method | Links | |||||
|---|---|---|---|---|---|---|
| ArcFaceTraining Dataset=VGGFace2, Backbone=ResNet-502022.05 | 99.19 | 92.07 | 97.8 | 99.67 | 99.73 | |
| SFaceTraining Dataset=VGGFace2, Backbone=ResNet-502022.05 | 99.19 | 92.51 | 98.19 | 99.68 | 99.8 | |
| SFaceTraining Dataset=MS1MV2, Backbone=ResNet-1002022.05 | 98.93 | 94.84 | 98.5 | 99.44 | 99.55 | |
| ArcFaceTraining Dataset=MS1MV2, Backbone=ResNet-1002022.05 | 98.83 | 93.47 | 98.11 | 99.33 | 99.51 | |
| SETraining dataset=MS+VF2, D=20482018.10 | 98.2 | 88.3 | 94.6 | 99.3 | 99.4 | |
| VGGFace22022.05 | 98.2 | 88.3 | 94.6 | 99.3 | 99.4 | |
| SETraining dataset=VF2, D=20482018.10 | 98.1 | 84.7 | 93 | 99.4 | 99.6 | |
| SE-GV-3Training dataset=VF2, D=1282018.10 | 97.9 | 87.2 | 95.1 | 99 | 99.2 | |
| UniformFace2022.05 | 97.9 | — | — | 98.8 | — | |
| SE-GV-4-g1Training dataset=VF2, D=1282018.10 | 97.7 | 88.4 | 95.1 | 99.1 | 99.4 | |
| L2-Face2022.05 | 97.3 | 91.5 | 95.6 | — | 98.8 | |
| Crystal Loss2022.05 | 97.2 | 91.8 | 95.9 | — | 98.8 | |
| FTL2022.05 | 96 | — | — | 98.3 | 98.7 | |
| NANTraining dataset=priv, D=1282018.10 | 95.8 | 81.7 | 91.7 | 98 | 98.6 | |
| NAN2022.05 | 95.8 | 81.7 | 91.7 | 98 | 98.6 | |
| DREAMTraining dataset=MS2018.10 | 94.6 | — | — | 96.8 | — | |
| Template Adaption2022.05 | 92.8 | 77.4 | 88.2 | 97.7 | 98.6 | |
| VGGFace2022.05 | 92.5 | 45.4 | 74.8 | 97.2 | 98.3 | |
| BinTraining dataset=ImNet+CAS, D=40962018.10 | 84.6 | 87.5 | — | 93.3 | 95.1 |